RADAR is a joint venture between data journalism start-up Urbs Media and the Press Association. Set up in September 2017 with funding support from Google’s Digital News Initiative, RADAR aims to find new ways of delivering quality local content at scale whilst also helping reporters and machines work together effectively.
Having recently attended the Press Association’s presentation on RADAR at the PPA Festival in London – you can download CEO Pete Clifton’s presentation here – WNIP caught up with RADAR’s Editor-in-chief, Gary Rogers, to find out more…
What business problem is your company addressing?
The major issue we’re tackling is “How can we scale the production of news to help financially stressed local newsrooms?” And, just as importantly, “How can we do so in a way that engages and informs consumers by delivering a high-volume of incisive, fact-based news stories?“
We saw the growing supply of open data as a neglected source of stories. Neglected because journalists to not have the time, or in some cases, the skills to dig through the data to find stories.
There is open data across all the main beats of news – health, crime, transport, education, housing, the environment – filled with stories waiting to be told. Much of this data is very granular with figures available at local level. It’s possible to write stories from this data for every local community in the UK.
What is your core solution addressing this problem?
Writing all those stories would either take too much time or too many people. We solved this problem by combining human journalism with automation. RADAR reporters work like other data journalists in digging out stories, but they write the story in the form of a template in Natural Language Generation software. NLG is one type of technology in the broad AI space. The data is fed through the template to produce multiple versions of the story.
Can you give examples of publishers successfully using RADAR?
We have a team of five reporters, publishing 8000-10,000 local stories each month. We work as a news agency. Our subscribers are UK local news publishers, among them big organisations like JPI, established brands like the Wolverhampton Express and Star, the UKRD radio network, right down to the hyperlocal Caerphilly Observer.
Stories are delivered via a subscriber web portal and published on local news websites across the UK and used extensively in print too.
RADAR is about local news, so our business model is based on geography. Our story distribution is divided into 391 local channels (mirroring the 391 local authority areas of the UK). Subscribers purchase access to one or more area, depending on their publishing footprint.
What are other people doing in the space and why?
Automation has been much discussed in the media and that conversation has been getting louder as companies look at how they make their content production more efficient. Many examples to date have looked at building a one-off template to deal with recurring data. There are several examples of sports results and financial reports bots.
RADAR is different because we work more dynamically. We have placed the technology in the hands of the journalists who create the content. Our team is writing 10 templates each week to deal with the vast array of available data.
How do you view the future?
AI is changing many industries. In publishing, it appears that the focus has been around serving content rather than creating it. That is going to change. There’ll be growing pressure to automate tasks. Part of the challenge is identifying the right tasks to automate. Machines are great at some things. Humans are still better at many. RADAR has shown the potential of harnessing human creativity with automated production. We are using technology to amplify the work of our reporters without compromising on the narrative skills.
The clue to this combination is in our name – Reporters and Data and Robots – RADAR.
Anything else you think we should know about?
Local news has been a fantastic proving ground for RADAR. What works for news can work in other types of content that can be driven by data.
Our news service is based on geography, but our methods could be applied to other forms of granularity – sector verticals, personalisation, etc.